May September 2007Sept 2007 Doc.: IEEE 802.22-07/0333R1 Doc.: IEEE 802.22-07/0Xxxr0 0
Total Page：16
File Type：pdf, Size：1020Kb
May September 2007Sept 2007 doc.: IEEE 802.22-07/0333r10
IEEE P802.22 Wireless RANs
[Comments on 802.22 quiet period scheduling with respect to the detection of 802.22.1 beacon MAC subframes]
Date: 2007-09-27
Author(s): Name Company Address Phone email David d.mazzarese@sa Samsung Electronics Korea +82 10 3279 5210 Mazzarese msung.com Samsung Telecom [email protected]. Baowei Ji USA +1-972-761-7167 America com +86 10 6439 0088 jinxia.cheng@sa Jinxia Cheng Samsung Electronics China 3133 msung.com cheng.shan@sam Shan Cheng Samsung Electronics Korea +82 31 279 7557 sung.com et.lim@samsung. Euntaek Lim Samsung Electronics Korea +82 31 279 5917 com Abstract [802.22 This document provides thea detailed analysis of the timing for detecting TG1 beacons, and of the disadvantages of is considering limiting the scheduling of a long quiet period to fall at the end of a superframe. ItThis could significantly delay the scheduling of inter-frame sensing quiet period, thus allowing continuous interference from the WRAN side to the wireless microphone even after a TG1 synch burst has been detected by the WRAN, in the event that the WRAN has not moved channels directly after detecting the TG1 sync burst. It could result in a large amount of waste idle time becauseif This would not allow a long quiet period cannot to start and end anywhere within a superframe. This analysis shows that the current length of the beacon in [1] is acceptable and allows for proper operation of the WRAN. The selected parameters in consideration of the design tradeoff could lead to unnecessarily
Notice: This document has been prepared to assist IEEE 802.22. It is offered as a basis for discussion and is not binding on the contributing individual(s) or organization(s). The material in this document is subject to change in form and content after further study. The contributor(s) reserve(s) the right to add, amend or withdraw material contained herein.
Release: The contributor grants a free, irrevocable license to the IEEE to incorporate material contained in this contribution, and any modifications thereof, in the creation of an IEEE Standards publication; to copyright in the IEEE’s name any IEEE Standards publication even though it may include portions of this contribution; and at the IEEE’s sole discretion to permit others to reproduce in whole or in part the resulting IEEE Standards publication. The contributor also acknowledges and accepts that this contribution may be made public by IEEE 802.22.
Patent Policy and Procedures: The contributor is familiar with the IEEE 802 Patent Policy and Procedures
Submission page 1David Mazzarese, Samsung ElectronicsDavid MazzareseDavid Mazzarese, Samsung Electronics May September 2007Sept 2007 doc.: IEEE 802.22-07/0333r10
1. Summary of 802.22.1 beacon structure
A motion was made during the plenary meeting in March 2007 in order to fix any inter-frame sensing periods at the end of superframes. The motion failed. However, there are still discussions on the locations of inter-frame sensing periods. For the benefits of the WG, this document provides the detailed analyses of the disadvantages of limiting those long quiet periods at the end of superframes.
The 802.22.1 Task Group has been making great efforts to design a beacon that is partitioned into 3 MAC subframes in order to allow an unlicensed device to capture parts of the beacon as desired, in order to limit both the latency and the amount of quiet time required to capture parts of the beacon.
Particular attention was paid to the detection of the beacon by the WRAN CPE and BS. The current superframe length (160 ms), frame length (10 ms) and SCH length plus propagation delay (2 ms) were taken into account in the analysis that lead to the final choice of PHY and MAC parameters in 802.22.1 [1]. The beacon was designed such that it can be captured in its entirety without interrupting the SCH with a maximum latency of 2 superframes from the start of the superframe right following the detection of a sync burst, provided that the WRAN can schedule the appropriate length quiet period to start and end anywhere within a superframe (synchronously with the start and end of PHY frames).
802.22.1 also paid particular attention to the length of the beacon MAC subframes, such that MSF1 fits within 30 ms, which would require at most a 40 ms quiet period and minimize the impact , thus still enablingto the continuous support of VoIP, which was pointed out to be one of the most critical parameters to consider.
In summary of the 802.22.1 parameters, the WRAN would need to capture either, or sequentially: - 0.1041 ms 8-chip complex PN spreading sequence within any quiet period (i.e. opportunistically) - 3.3302 ms sync burst within a period (short) quiet period, which takes no longer than 5 ms asynchronously - 29.1396 ms MFS1 (including FEC, but no authentication) - 70.7676 ms MSF1_FEC+MSF2 (allowing for authentication at the BS, but not at the CPE) - 99.9072 ms full beacon (PHY_header+MSF1_FEC+MSF2+MSF3)
The period of the 802.22.1 beacon is 103.2374, which includes an additional 3.3302 ms slot for RTS/ANP burst between 2 beacon frames. The most critical part to receive is the MSF1, since the sync burst is repeated continuously (with period 3.3302 ms). In the sequel, a long quiet period refers to a quiet period longer than 1 WRAN PHY frame (10 ms).
2. Qualitative Analysis
There are a few effects to consider together in order to fully understand the impact of the type of scheduling of the long quiet period:
. the minimum required quiet period length,
. the statistics of the latency before being able to receive the desired part of the 802.22.1 beacon (including the superframe in which it is actually detected) from the start of the superframe right following the detection of a sync burtburst, assuming asynchronicity between the 802.22 superframe and the 802.22.1 frame,
. the maximum detection time assuming some detection errors, and re-attempts to detect the beacon,
. the wasted transmission time: the time during a quiet period where the WRAN doesn’t need to listen to the 802.22.1 beacon, whenever the quiet period is longer than the required quiet period length, and the unavoidable waster due to the asynchronicity of the start of the WRAN PHY frame and the beacon frame. Submission page 2David Mazzarese, Samsung ElectronicsDavid MazzareseDavid Mazzarese, Samsung Electronics May September 2007Sept 2007 doc.: IEEE 802.22-07/0333r10
In order to meet the 2 seconds functional requirements to detect an incumbent and vacate the channel, the maximum detection time assuming only one attempt at detection the beacon should be below 1 second, in order to allow enough time for a second attemps , and so that enough time is left for all those CPEs that have detected the TG1 beacons to report back to the BS, and for the BS and the whole cell to finish channel moving timely. Thus, the maximum latency should be kept below 1 second, i.e., 6.2512.5 superframes. It should be noted that if the WRAN vacates the channel without decoding the information contained in the beacon, it could be wasting some resources and also in the case of multiple overlapping WRANs, there would be no possibility for one WRAN to alert another WRAN about the location of the Part 74 device to be protected.
The impact of scheduling a long quiet period to always end at the end of a superframe is mostly seen on the wasted transmission time. It is possible that the MSF1 part of the beacon falls towards the beginning of the superframe, which would require a quiet period much longer than the 30 ms required to detect MSF1.
An alternative to alleviate this problem is to wait for a few more superframes, and only schedule the long quiet period when the MSF1 part (or any desired part) of the beacon falls within the last L PHY frames of a superframe. For example, it was suggested to wait until the MSF1 falls within the last 4 or 5 PHY frames for scheduling the long quiet period. The impact of this strategy is to increase the statistics of the latency, and accordingly impact the maximum detection time. Note that the value of L will need to be defined for each possible length of 802.22.1 beacon parts to detect at the WRAN BS and CPE.
An analysis of all the above suggested scenarios is provided in the remainder of this document. It is based on the mathematical formulation of the asynchronous nature of the 802.22 and 802.22.1 devices, as detailed below.
As long as the quiet period length is smaller than one superframe length (158 ms without SCH), the WRAN does not need to interrupt the superframe preamble to schedule a quiet period, but it needs to wait until the desired part of the beacon frame does not interrupt the superframe preamble.
S H
SCH WRAN Superframe SCH WRAN Superframe
Beacon Frame Beacon Frame Beacon Frame D
B Quiet period P
If D+P < S, the quiet period can be scheduled in the current superframe (it is assumed that B < S). Otherwise the minimum number of superframes to wait for before scheduling a quiet period is:
mS H D arg minm 1 S B B P 0 m B
Submission page 3David Mazzarese, Samsung ElectronicsDavid MazzareseDavid Mazzarese, Samsung Electronics May September 2007Sept 2007 doc.: IEEE 802.22-07/0333r10
3. Numerical Analysis
Detection of MSF1
Let us first focus on the detection of the MSF1 part of the beacon. We set the maximum latency to 320 superframes (51.2 seconds) for the sake of observing all the statistics. The results are summarized in Table 1, while the full statistics are available in the following figures. The random intialinitial offset (in ms) represents the amoungamount of random asynchronicity between the 802.22 superframe and the 802.22.1 beacon. max number (L) of 3 (30 ms) 4 (40 ms) 5 (50 ms) 6 (60 ms) 16 (158 ms) quiet frames at the no constraint end of a superframe (current draft) length of required 29.139600 ms 29.139600 ms 29.139600 ms 29.139600 ms 29.139600 ms quiet period (MSF1) (MSF1) (MSF1) (MSF1) (MSF1) mean latency before detection is 1043 ms 611 ms 412 ms 32 ms scheduling a long impossible within (7 superframes) (4 superframes) (3 superframes) (< 1 superframe) period 51.2 seconds max latency before detection is 3200 ms 1440 ms 1120 ms 160 ms scheduling a long impossible within (20 superframes, (9 superframes) (7 superframes) (1 superframe) period 51.2 seconds unacceptable) average wasted N/A 4.560174 ms 6.802656 ms 10.406132 ms 60.777210 ms quiet time during quiet period (if quiet until the end of the superframe)
Table 1: Constraints on detecting the 802.22.1 beacon field MSF1
As shown in Table 1, there are many scenarios that make it impossible to use a quiet period of 3 or 4 quiet frames at the end of superframes to detect MSF1 within 2 seconds. So L must be larger than 4. However, for L equals 5 or 6, the average latency before scheduling a long quiet period is quite big, which implies that wireless microphones hasve to accept the continuous interference from WRAN for 412ms to 611ms on average even though the latter has detected the synch burst of the Part74 Beacon Devices. The maximum latency before scheduling a long quiet period is from 1120 ms to 1440 ms. This is all because that the WRAN could not schedule the inter-frame sensing period to starting in other places exceptother thatn of 5 or 6 frames before the end of the superframe.
While the latency is reduced if longer quiet period is used, the wasted time is increased. The last column shows that the quiet period of 158 ms has delays of 32ms on average, but the averaged waste time is more than 60 ms. The highlighted value (60.777210 ms) can be saved if the quiet period can be scheduled to start at any time and end at any time within a superframe. It is noticeable that this average wasted quiet time is about twice as long as the actual required quiet period, which means that on average, the WRAN will schedule a quiet period 3 times as long as required, if it needs to schedule the quiet period until the end of the superframe.
By waiting to schedule the quiet period until it falls within the last L frames of a superframe, this average wasted quiet time can be saved, but the maximum latency increases from 160 ms to more than 1 second, which only allows for one attempt to decode the beacon. If it fails, the WRAN will have to vacate the channel without knowing the information contained in the beacon, but based solely on its earlier detection of the beacon sync burst. With 6 quiet frames at the end of the superframe, ti it still takes more than 1 second of latency in more than 10% of cases.
Submission page 4David Mazzarese, Samsung ElectronicsDavid MazzareseDavid Mazzarese, Samsung Electronics May September 2007Sept 2007 doc.: IEEE 802.22-07/0333r10
Detection of MSF1+MSF2
Let us now turn our attention on the detection of the MSF1+MSF2 parts of the beacon for authentication at the base station. Due to the length of the required quiet period, it does not make sense to wait beyond the first superframe where the MSF1+MSF2 parts would be detectable without interrupting the SCH. Therefore, if we refer to the last column in Table 2 below, we see that on average, we would need to schedule a quiet period of length 110 ms if it ends at the end of a superframe, thus wasting an average 38.18 ms of transmission time whenever an 802.22.1 beacon is present. max number (L) of 8 (80 ms) 9 (90 ms) 10 (100 ms) 11 (110 ms) 16 (158 ms) quiet frames at the no constraint end of a superframe (current draft) length of required 71.600160 ms 71.600160 ms 71.600160 ms 71.600160 ms 71.600160 ms quiet period (MSF1+MSF2) (MSF1+MSF2) (MSF1+MSF2) (MSF1+MSF2) (MSF1+MSF2) mean latency before 1357 ms 596 ms 390 ms 240 ms 88 ms scheduling a long (9 superframes) (4 superframes) (3 superframes) (2 superframes) (1 superframe) period max latency before 3040 ms 1280 ms 960 ms 800 ms 320 ms scheduling a long (19 superframes, (8 superframes) (6 superframes) (5 superframes) (2 superframes) period unacceptable) average wasted 3.165632 ms 6.122688 ms 9.178504 ms 14.034664 ms 38.183546 ms quiet time during quiet period (if quiet until the end of the superframe)
Table 2: Constraints on detecting the 802.22.1 beacon fields MSF1+MSF2
Detection of MSF1+MSF2+MSF3 (full beacon)
Let us now turn our attention on the detection of the full beacon (99.9072 ms) for authentication at the CPE. As can be seen in the last column of Table 3 below, it is not a big constraint to schedule the long quiet period at the end of the superframe, since that quiet period will interrupt the WRAN for 100 ms in any case. max number (L) of 10 (100 ms) 11 (110 ms) 12 (120 ms) 13 (139 ms) 16 (158 ms) quiet frames at the no constraint end of a superframe (current draft) length of required 99.9072 ms 99.9072 ms 99.9072 ms 99.9072 ms 99.9072 ms quiet period (full beacon) (full beacon) (full beacon) (full beacon) (full beacon) mean latency before detection is 112 ms 590 ms 392 ms 145 ms scheduling a long impossible within (8 superframes) (4 superframes) (3 superframes) (1 superframe) period 51.2 seconds max latency before detection is 3200 ms 1440 ms 1120 ms 320 ms scheduling a long impossible within (20 superframes, (9 superframes) (7 superframes) (2 superframes) period 51.2 seconds unacceptable) average wasted N/A 4.114064 ms 6.658284 ms 9.971292 ms 30.098168 ms quiet time during quiet period (if quiet until the end of the superframe)
Table 3: Constraints on detecting the whole 802.22.1 beacon frame
Submission page 5David Mazzarese, Samsung ElectronicsDavid MazzareseDavid Mazzarese, Samsung Electronics May September 2007Sept 2007 doc.: IEEE 802.22-07/0333r10
4. Conclusions
It was suggested to define a long (non-periodic) quiet period by its start time within a superframe, and to let it be last until the end of the superframe. This document provided an analysis of this type of scheduling of a long quiet period from the point-of-view of the detection of an 802.22.1 beacon.
The emphasis on the detection of the MSF1 part of the 802.22.1 beacon showed that:
. An average 90 ms of quiet period would need to be scheduled to detect a 30 ms MSF1 field, thus disabling the ability to support VoIP during sensing (whereas 802.22.1 has been striving to meet that constraint). This can be avoided by allowing the quiet period to start and end at any time within a superframe.
. Constraining the long quiet period to fall within the last 5 or 6 frame of a superframe would incur a maximum latency of more than 1.1 second before meeting that constraint, thus disabling any possibility to re- attempt detection of MSF1 in the event that the first detection failed, due to the 2 seconds constraint to leave the channel after the detection of an incumbent.
. If such a constraint was defined to limit the amount of wasted quiet time, given that the quiet period is always scheduled at the end of a superframe, then it should be defined for every possible length of 802.22.1 packet that the WRAN would have to detect, and this should be defined in the MAC of 802.22, or left as an implementation issue and increased complexity for the operator. A more complex timer is would also be required, rather than to simply allow for a long quiet period anywhere in any superframe (without interrupting the SCH, which only requires to skip one superframe if needed).
. The situation where multiple WRANs need to detect a beacon simultaneously, or rather consecutively, detect a beacon has not been analyzed, and would put further constraints on the latency.
. The impact of constraining the long quiet period to end at the end of a superframe is not significant for the detection of the whole beacon, or MSF1+MSF2 parts, since in both cases VoIP cannot be supported due to the minimum length of the quiet period.
. Note that if continuing support of VoIP is provided during the time when a WRAN switches its operation to another channel, then continuing support of VoIP must also be provided during any type of sensing (periodic or not) as long as the WRAN is sensing on-channel. This also suggests that sensing for detecting the 802.22.1 MSF1+MSF2 or the whole beacon frame would need to be done after vacating the channel.
. There is no impact on the synchronization of the quiet periods among adjacent cells, since the scheduling of a long quiet period is not periodic, and should be performed as a response to the detection of an 802.22.1 sync burst. There is no need and no possibility to synchronize these quiet periods. The first WRAN to be able to decode an 802.22.1 beacon frame (the closest one to the beacon) should first vacate the channel and then alert the neighbouring WRANs with the information contained in the beacon.
The authors believe it does not make good engineering sense to limit a long quiet period to always end at the end of a superframe. By allowing a long quiet period to end before the end of the superframe, the 802.22 standard will make its best effort to protect the incumbents according to the functional requirements, as required in the PAR, and as stated by the joint proposal group in March 2006, and it will only incur a marginal increase in signalling overhead.
The analysis in this document is consistent with the motion result. The parameters in the current WG draft current [2] also support the scheduling of inter-frame sensing periods starting and ending anywhere inside a superframe.
Submission page 6David Mazzarese, Samsung ElectronicsDavid MazzareseDavid Mazzarese, Samsung Electronics May September 2007Sept 2007 doc.: IEEE 802.22-07/0333r10
We propose to modify the MAC message to allow the BS to signal the end of a long quiet period explicitely (e.g. for long quiet periods), while the default mode of scheduling a quiet period (e.g. for periodic quiet frames) would be such that the quiet period ends at the end of a superframe. This is not an optional feature. The default mode can be interpreted from the structure of the message. If the end of the quiet period is not explicitely stated in the message, then the quiet period will end at the end of the superframe, otherwise it will end as signalled by the BS.
[Baowei: please include the change of MAC message required for that]
5. References
[1] P802.22.1/D1, Part 22.1: Enhanced Protection for Low-Power, Licensed Devices Operating in Television Broadcast Bands, May, 2007.
[2] IEEE P802.22™/D0.3.7: Draft Standard for Wireless Regional Area Networks Part 22: Cognitive Wireless RAN Medium Access Control (MAC) and Physical Layer (PHY) specifications: Policies and procedures for operation in the TV Bands, July 2007.
6. Appendix
In the figures below, the red lines represent the 2 seconds limit for the detection of an incumbent.
Figures 1 to 4 show the statistics of the timing parameters for the detection of MSF1 as a function of the random initial timing offset between the WRAN and TG1 superframes, when the sensing window size at the end of a superframe is L=4, 5, 6 and 16 frames.
Figure 5 shows the statistics of the timing parameters for the detection of the whole beacon frame as a function of the random initial timing offset between the WRAN and TG1 superframes, when there is no constraint on the size of the sensing window (16 WRAN frames).
Submission page 7David Mazzarese, Samsung ElectronicsDavid MazzareseDavid Mazzarese, Samsung Electronics May September 2007Sept 2007 doc.: IEEE 802.22-07/0333r10
Figure 1: detection of MSF1 with L=4 frames
Submission page 8David Mazzarese, Samsung ElectronicsDavid MazzareseDavid Mazzarese, Samsung Electronics May September 2007Sept 2007 doc.: IEEE 802.22-07/0333r10
Figure 2: detection of MSF1 with L=5 frames
Submission page 9David Mazzarese, Samsung ElectronicsDavid MazzareseDavid Mazzarese, Samsung Electronics May September 2007Sept 2007 doc.: IEEE 802.22-07/0333r10
Submission page 10David Mazzarese, Samsung ElectronicsDavid MazzareseDavid Mazzarese, Samsung Electronics May September 2007Sept 2007 doc.: IEEE 802.22-07/0333r10
Figure 3: detection of MSF1 with L=6 frames
Submission page 11David Mazzarese, Samsung ElectronicsDavid MazzareseDavid Mazzarese, Samsung Electronics May September 2007Sept 2007 doc.: IEEE 802.22-07/0333r10
Submission page 12David Mazzarese, Samsung ElectronicsDavid MazzareseDavid Mazzarese, Samsung Electronics May September 2007Sept 2007 doc.: IEEE 802.22-07/0333r10
Figure 4: detection of MSF1 with L=16 frames
Submission page 13David Mazzarese, Samsung ElectronicsDavid MazzareseDavid Mazzarese, Samsung Electronics May September 2007Sept 2007 doc.: IEEE 802.22-07/0333r10
Submission page 14David Mazzarese, Samsung ElectronicsDavid MazzareseDavid Mazzarese, Samsung Electronics May September 2007Sept 2007 doc.: IEEE 802.22-07/0333r10
Figure 5: detection of MSF1+MSF2+MSF3 with L=16 frames
Submission page 15David Mazzarese, Samsung ElectronicsDavid MazzareseDavid Mazzarese, Samsung Electronics